Portail HAL ONERA
Not a member yet
12842 research outputs found
Sort by
Développement et validation d'une stratégie de maillage pour des structures fabriqués par placement automatisé de fibres
No full textInternational audienceThe Automated Fiber Placement (AFP) process is known for its design flexibility, enabling the design and manufactureof parts with non-conventional lay-ups. Composite structures featuring variable fiber angle is one of the structural casesof interest in the aerospace industry, given their performance potential. Nevertheless, manufacturing structures withAFP generates singularities within the layers. The impact of these discontinuities on the mechanical behavior anddesign of such structures remains poorly understood. This work presents a meshing strategy based on geometry and tapedeposition path. This approach generates an idealized mesh of the geometry post-robot deposition, as thermomechanicaleffects of the process are not accounted for at this stage. Finally, the validation of the method is performed throughcomparison with a high fidelity mesh generated from micro-tomography scan segmentation.Le procédé de placement automatisé de fibres (AFP) est connu par sa flexibilité de design, ce qui permet la conceptionet la fabrication de pièces avec une dépose non conventionnelle. Les structures composites contenant des fibrescurvilignes pour la réalisation de panneaux à rigidité variable (VSP en anglais) suscitent un intérêt croissant dansl’industrie aéronautique par leur potentiel de performance. Cependant, la fabrication de structures par AFP peutengendrer des singularités au sein des plis. L’impact de ces discontinuités sur le comportement mécanique et le designde ces structures reste peu connu. Ce travail présente une stratégie de maillage fondée sur une description géométriqueanalytique des chemins de dépose de bandes. Cette approche génère un maillage idéalisé de la géométrie après ladépose du robot puisque les effets thermomécaniques du procédé ne sont pas pris en compte. Enfin, la validation de laméthode est réalisée à partir d’une comparaison fine entre la géométrie générée par celle dernière et celle générée d’unesegmentation de données issues d’observations microtomographiques
Etude de la fracturation du Permien Rouge du bassin de Lodeve : mise en avant de nouveaux evenements de remobilisations de metaux
No full textNational audienc
A Diffeomorphic Mapping Approach for Model Order Reduction in Aerodynamics
No full textInternational audienc
Autoencodeurs variationnels à valeurs complexes pour la détection radar dans un fouillis gaussien mixte et un bruit thermique
No full textInternational audienceThis paper introduces a novel Complex Variational AutoEncoder architecture leveraging complex-valued representations with a single latent channel for radar target detection. Unlike conventional real-valued VAEs, which require two latent channels to encode complex-valued data, the proposed approach directly operates in the complex domain, preserving the inherent structure of radar signals. By maintaining phase information and more accurately modeling the underlying data distribution, the complex-valued VAE enhances the separation of latent representations, leading to improved target discrimination.Experimental results demonstrate the effectiveness of the proposed CVAE in various clutter and noise scenarios against traditional radar detectors, such as the Matched Filter and Adaptive Normalized Matched Filter, and real-valued VAE.Cet article présente une nouvelle architecture d'autoencodeur variationnel complexe tirant parti de représentations à valeurs complexes avec un seul canal latent pour la détection de cibles radar. Contrairement aux VAE conventionnels à valeurs réelles, qui nécessitent deux canaux latents pour coder des données à valeurs complexes, l'approche proposée fonctionne directement dans le domaine complexe, en préservant la structure inhérente des signaux radar. En conservant les informations de phase et en modélisant plus précisément la distribution des données sous-jacentes, la VAE à valeurs complexes améliore la séparation des représentations latentes, ce qui permet d'améliorer la discrimination des cibles. Les résultats expérimentaux démontrent l'efficacité de la CVAE proposée dans divers scénarios de fouillis et de bruit par rapport aux détecteurs radar traditionnels, tels que le filtre adapté et le filtre adapté normalisé adaptatif, ainsi que la VAE à valeurs réelles
EMC Analysis on the Integration of an Electric Propulsion Architecture
No full textInternational audienceThe aim of this paper is to analyze the impact of constraints linked to the integration of an electrical power architecture at system level. Based on the description of a process used to define and evaluate the characteristics of an electrical architecture, the multidisciplinary effects of the various integration constraints (electromagnetic, thermal, etc.) linked to the power increase are identified. Based on these findings, a multiphysics pre-analysis is carried out to assess the effects of the installation choices on its electromagnetic response
An Optimised Quasi-Trefftz Method for the Iterative Solution of Time-Harmonic Wave problems
No full textInternational audienc
Wideband 1-bit Reconfigurable Circularly Polarized Transmitarrays in Ka-Band
No full textInternational audienceThis study details the design of an active Transmitarray (TA) operating in the Ka-band made of a reconfigurable linear-to-circular-polarization (LP-CP) unit-cell. The unit-cells incorporate a true-time-delay (TTD) polarization conversion mechanism and a stacked patch configuration to achieve wide transmission and axial ratio bandwidths. Each cell integrates a pair of PIN diodes for electronically controllable 180° phase shifts. This is the first demonstration of a wide-band reconfigurable and circularly polarized unitcell in millimeter wave bands, without the use of an external polarizer. A 20 × 20 cells Ka-band circularly polarized TA was designed and fabricated. A random symmetric distribution of rotations applied to the inner patch antenna enhances CP purity and increases the scan range. Measurement results demonstrate a broadside gain of 20.55 dB, with both -3 dB gain bandwidth and 3 dB axial ratio bandwidth exceeding 22%, which surpasses all previous millimeter-wave reconfigurable design. In the case of the 20 × 20 cell array manufactured the scan range extends to ±60 • in both E and H planes, with a maximum scan loss of 4.2 dB and axial ratio values under 2 dB for elevation angles up to ±50°. Additional full wave simulations of a larger 60 × 60 cell array demonstrate a better maximum scan loss of 2.4 dB for elevation angles reaching ±50° , 3.8 dB for an elevation angle of ±60 • and 6 dB for an elevation angle of ±70 • with axial ratio values consistently under 3 dB
Exploring the formation of multiple dust-trapping rings in the inner Solar System
No full textInternational audienceContext. Isotopic properties of meteorites provide evidence that multiple dust trap or pressure bumps had to form and persist in the inner Solar System on a timescale of millions of years. The formation of a pressure bump at the outer edge of the gap opened by Jupiter would be effective in blocking particles drifting from the outer to the inner disc. Yet this would not be enough to preserve dust in the inner disc. However, in low-viscosity discs and under specific conditions governing the gas cooling time, it has been shown that massive planets can also open secondary gaps, separated by density bumps, inwards of the main gap. The majority of studies of the process of secondary gap formation have been done in two dimensional equatorial simulations with prescribed disc cooling or by approximating the cooling from the disc photosphere. Recent results have shown that an appropriate computation of the disc cooling by including the treatment of radiation transport is key to determining the formation of secondary gaps.Aims. Our aim is to extend previous studies to three dimensional discs by also including radiative effects. Moreover, we also consider non-ideal magnetohydrodynamic effects in discs with a prescribed cooling time to explore the feedback of the magnetic field on secondary gap formation.Methods. We performed three dimensional hydrodynamical numerical simulations with a self-consistent treatment of radiative effects making use of a flux-limited diffusion approximation. We then extended our study to a similar disc including the magnetic field and non-ideal Ohmic and ambipolar effects.Results. We show that in the hydrodynamical model, a disc with low bulk viscosity (αν = 10−4) and consistent treatment of radiative effects, a Jupiter-mass planet is capable of opening multiple gaps. We also show that multiple gaps and rings are formed by planetary masses close to the pebble isolation mass. In the presence of non-ideal MHD effects, multiple gaps and rings are also formed by a Jupiter-mass planet.Conclusions. A solid Jupiter core in low-viscosity discs blocks particles drifting towards and within the inner disc. The formation of multiple gaps and rings inside the planetary orbit at this stage is crucial to preserving dust reservoirs. Such reservoirs are pushed towards the inner part of the disc during Jupiter runaway growth and they are shown to be persistent after Jupiter’s growth. Multiple dust reservoirs could therefore be present in the inner Solar System since the formation of Jupiter’s solid core when the disc is characterised by low viscosity
A Novel Microwave Resonant Probe Diagnostic to Perform Real-Time and In-Situ Electron Density Measurements in Atmospheric Re-Entry Plasmas
No full textInternational audienceIn this work, the design, manufacturing, and validation of a new high-frequency curling probe operating near 20 GHz is presented. This probe is capable of accessing electron densities in the 10 12 -10 14 cm -3 range, typical of atmospheric re-entry plasmas. A new interpretation model, accounting for collisional effects in high-density plasmas, is developed and validated through 3D electromagnetic simulations of the probe frequency response. For the first time, the curling probe diagnostic is deployed in a highenthalpy supersonic plasma wind tunnel (PHEDRA facility at ICARE Laboratory in Orléans, France) to characterize electron density across the shock layer of a coin-shaped model under re-entry relevant conditions. Boron nitride protective coatings have allowed the probes to withstand cumulative plasma exposure of 30 minutes under thermal fluxes of 200 -300 kW/m² and gas temperatures of thousand kelvins. Curling probe measurements provided integral electron densities across several millimeters of the shock layer, yielding values on the order of 10 13 cm -3 . These results are in fair to good agreement with cylindrical Langmuir probe results and consistent with complementary direct Monte Carlo simulations of the plasma flow. These findings demonstrate the feasibility and robustness of curling probe for diagnosing high-density plasmas in re-entry conditions, highlighting their potential for advanced ground-based experimental campaigns and for future in-flight applications.</div
Reachability Graph Learning with Reinforcement Learning Control for Robotic Tasks
No full textInternational audienceDeep reinforcement learning (RL) has proven to be an effective method for learning robot control in simulation, but it suffers from sample inefficiency and lacks generalization capabilities in real-world applications. These weaknesses are often due to unrealistic assumptions, including the availability of a perfectly accurate simulator, the ability to collect vast amounts of training data, and the existence of a well-defined, single reward function. Coupling learned control policies with global automated planning for skill composition has improved performance in both simulated and real-world robotics settings. In this paper, we propose an adaptation of the Reachability Graph Learning algorithm for robotic applications. Our algorithm is able to explore and build a reachability graph directly by learning in the real world, without resetting the agent state. By discarding unnecessary information, especially when dealing with large observation spaces, we demonstrate the ability to construct a low-cost control policy that is highly efficient for real-world robotic goal-reaching tasks